利用K均值算法改进后的蚁群优化算法对高光谱图像聚类研究  

Ant Colony Optimization Algorithm for Hyperspectral Image Clustering Based on K Means Algorithm Improved

作　　者：王秀和[1] 

机构地区：[1]防灾科技学院信息化管理中心,河北三河065201

出　　处：《科技通报》2015年第3期202-206,共5页Bulletin of Science and Technology

基　　金：国家高技术研究发展计划(863计划)专题(2012AA121304);(2012AA121304-3)

摘　　要：在对K均值算法和图像聚类的蚁群优化算法(ACO)进行对比分析后,本文提出了K均值ACO算法来解决K均值误分类和ACO慢聚敛的问题。K均值ACO算法将K均值的结果当成ACO的启发式信息,如此提高了ACO在蚁群寻找规则方面的光照概率和光照像素,允许蚁群根据信息素集中度而非概率来筛选节点,以致可以在不对ACO的随机搜索质量进行改变的情况下就可以完全获取到启发式信息。对模拟数据和真实数据进行验证后,K均值ACO算法证实可以改善K均值误分类的聚类精确度并提高ACO的收率速度。In the ant colony optimization algorithm on the K mean algorithm and image clustering （ACO）were compared and analyzed, this paper puts forward K mean ACO algorithm to solve the K mean errorclassification and ACO slow convergence problem. K means ACO algorithm will mean K results asheuristic information to improve the ACO, so ACO in ant colony search rules light probability and lightpixels, allowing the ant colony based on pheromone concentration rather than the probability to selectnodes, which can be changed randomly search quality is not ACO＇s case can be completely access to theheuristic information. To verify the simulation data and real data, the mean K ACO algorithm that canimprove the K mean clustering accuracy error classification and improve yield rate of ACO.

关 键 词：高光谱 K-MEANS 蚁群优化算法 聚类 

分 类 号：TP301.5[自动化与计算机技术—计算机系统结构]